Annex to Quirke et al. Quality assurance in pathology in colorectal cancer screening and diagnosis: annotations of colorectal lesions

The terminology is still under discussion. Serrated lesions can be regarded as a continuous spectrum of colorectal lesions with increasingly more pronounced serrated morphology starting with a hyperplastic polyp (HP) and progressing to sessile serrated lesions (SSLs, sometimes referred to as sessile serrated adenomas or sessile serrated polyps), traditional serrated adenomas (TSA) and leading, finally, to adenocarcinoma. Not only the adenomatous component but also other alterations associated with more pronounced serrated morphology may potentially progress to cancer (see Table 2).

The situation involving sessile serrated lesions is complicated as these lesions only reveal complex structural abnormalities, not adenomatous changes. Therefore, these lesions are neither adenomatous nor are they neoplastic. This is why Kudo et al. [9] and Lambert et al. [10] recommended that these lesions no longer be called adenomas; instead they should be referred to as SSLs. Few of these lesions are reported to rapidly progress to invasive carcinoma, [11]. Those few cases that do progress rapidly, particularly in the right colon, may be expected to appear more frequently as interval cancers. TSAs, unlike SSLs, do contain adenomatous alterations, albeit sometimes quite subtle [12]; they are therefore termed correctly and treatment and surveillance should correspond to that of adenomas (see Chapters 8 and 9 in the full guideline document [13, 14]).

The different histopathologic types (e.g. HP and SSL, SSL and TSA, adenoma and SSL, etc.) must be stated in the diagnosis.

Hyperplastic polyps (HPs) are composed of elongated crypts (no complex architecture) with serrated architecture in the upper half of the crypt. These polyps usually show some proliferation in the basal (non-serrated) part of the crypts (regular proliferation). Nuclei are small, regular, basal-orientated and lacking hyperchromasia, but with stratification of the upper (serrated) half of the crypts, and without cytological or structural signs of neoplasia.

Differences in the appearance of the cytoplasma permit recognition of three types:

The microvesicular variant greatly predominates, but distinction between types is subject to wide interobserver variation, especially in small lesions, and is not always possible. Currently, routine subclassification is therefore neither feasible, nor has it been shown to be beneficial.

At the molecular level the microvesicular variant of HP may be the precursor lesion for sessile serrated lesion, and a goblet-cell-rich HP may be the precursor lesion for a traditional serrated adenoma [15‐17]. Routine distinction of these types is not necessary.

Sessile serrated lesions are described in the literature as “sessile serrated adenoma” and are often found in the right colon. This is a misnomer since sessile serrated lesions do not contain adenomatous changes [9, 10, 18].

To date, four synonymously used terms exist for these lesions: sessile serrated adenoma [19], superficial serrated adenoma [20], type 1 serrated adenoma [21] and serrated polyp with abnormal proliferation [17].

We recommend using only the term sessile serrated lesion and avoiding use of any other terms for this entity. This recommendation is given in full awareness that sessile serrated lesions do not show histological signs of an adenoma, but, like adenomas, they should be excised if detected during an endoscopic examination. Currently, even in the hands of expert GI pathologists, the agreement on the sub-types of serrated lesions is only moderate [22].

The vast majority of SSLs will not progress to adenocarcinoma. Histological criteria of these sessile, usually larger lesions include an abnormal proliferation zone with structural distortion, usually most pronounced in dilatation of the crypts, particularly near the base. Abundant mucus production is usually also observed as pools of mucin in the lumen of the crypts and on the surface of the mucosa. SSLs are found mainly in the right colon and may be misdiagnosed as hyperplastic polyps. Clues to the correct diagnosis include location and large size. As discussed above, cytological signs of “neoplasia” are lacking, but structural abnormalities are present, i.e. glandular branching [18].

Sessile serrated lesions have an elevated serration index and serration in the basal half of crypts with basal dilation of crypts. The epithelium/stroma-ratio is believed to be >50% in SSL. There is crypt branching with horizontal growth (above muscularis mucosae; e.g. T- and L-shaped glands) and often pseudoinvasion into the submucosal layer, rectangular dilation of whole crypts with and without presence of mucus, increased number of goblet cells at the base of the crypts, vesicular nuclei with prominent nucleoli and proliferation zone in the middle of the crypts. Currently there is insufficient evidence available in the literature for weighting of these criteria.

A well-oriented polypectomy is mandatory for the identification of such histological features. Correct assessment of the deepest portions of the mucosa is impossible in superficial or tangentially cut lesions [15, 16].

Further criteria include an often asymmetrical expansion of the proliferation zone into the middle third of crypts. Often mild cytological atypia (slightly enlarged vesicular nuclei, nucleoli) is found without clear signs of neoplasia (dysplasia).

BRAF mutations depend on the type and location of lesion (see Table 3).

Other abnormalities include:

The frequency of sessile serrated lesions in small retrospective series is estimated at 2–11% of all mucosal lesions in the colon [35, 36]; between 8% and 23% are misdiagnosed as hyperplastic polyps with an interobserver variation of up to 40% [17, 37‐39].

The histological features separating HPs from SSLs constitute a continuous spectrum, and intermingled features can often be seen. This could explain the moderate interobserver concordance (k = 0.47) and the overlapping proliferative activity, and may justify establishing semi-quantitative criteria for diagnosis (e.g. >30% of undifferentiated cells) [40, 41]. Only a few immunohistochemical markers (Ki67, Ki67 + CK20, MUC6) have been tested for differentiating HPs and SSAs, and their usefulness in colorectal screening and diagnosis remains to be validated [42, 43]. At present, such an additional immunohistochemical analysis cannot be recommended (see Table 4).

In all likelihood, lesions formerly interpreted as mixed hyperplastic and adenomatous polyp are, in fact, SSLs complicated by conventional neoplasia, [44]. Special care must be taken in such cases to document the respective histopathologic components in such mixed polyps. Sometimes the conventional neoplastic part shows features other than in classical adenomas. The nuclei are prominent, less palisading and smaller than in classical adenomas. It is not clear whether this type of morphology is distinct for serrated lesions and whether any clinical implications can be drawn.

Prospective studies with risk stratification are needed to develop more precise methods of diagnosis and recommendations for classification. Sessile serrated lesions appear to take a long time (average 17 years) to develop into an invasive carcinoma. In contrast, an ill-defined, small subsample of SSLs seems to rapidly progress [11, 44]. Therefore, SSLs should be completely excised, particularly if they are located on the right side of the colon [16, 45].

Diagnosis on a biopsy is not adequate to exclude SSL since the most severe histologic changes might only appear focally within a lesion that otherwise appears to be a hyperplastic polyp [46].

The German guidelines for colorectal cancer [47] recommend complete removal and follow-up of SSL similar to adenomas. An intensive surveillance protocol is recommended for sessile serrated lesions (surveillance colonoscopy after 3–5 years subsequent to complete excision of non-neoplastic SSL, after 1 year after excision of SSL HGIEN [47]).

The UK Guidelines [48‐51] recommend complete excision but classify these lesions in the same risk category as hyperplastic polyps. The existing evidence base is not definitive as to the level of risk, and follow-up decisions should be made locally until more evidence is forthcoming.

Traditional serrated adenomas show neoplastic crypts with a serrated structure [52]. Compared to hyperplastic polyps, the most striking diagnostic feature of traditional serrated adenomas are the complex serrated morphology and the eosinophilic, “dysplastic” cytoplasm that still can be identified in cases with invasive adenocarcinoma. These lesions also frequently show BRAF mutations and CIMP with hMLH1 promoter methylation. Additionally, so-called intraepithelial microacini can be observed in the upper half of the mucosa (ectopic crypt formation). Often, these lesions are located in the distal colon and can be found more frequently in elderly female individuals [12, 18, 43].

A mixed polyp may contain partially hyperplastic, classical adenomatous or traditional serrated adenoma or sessile serrated lesion components. Rather than a continuous spectrum such lesions most probably represent several evolutionary lines, depending on the order of certain abnormalities in genes such as APC, BRAF and KRAS [15, 16]. It has to be determined whether mixed polyps represent serrated lesions complicated by conventional neoplasia [53].

Focal, hyperplastic-like narrowing of the basal region of a few crypts in SSL and the findings of flat sectors or ectopic crypt formation in SSL/TSA [43] are examples of combinations of serrated and adenomatous components. However, these features add no information of further diagnostic value; they probably result from the continuous developing nature of serrated lesions. We therefore recommend that the diagnosis of mixed polyp should be restricted to the definition given above under Terminology (i.e. lesions with—…more than one histopathologic type in the serrated spectrum (HP, SSL, TSA) … or at least one type in combination with adenomatous tissue…). Mixed polyps are serrated lesions in which more than one histopathologic type in the serrated spectrum (HP, SSL, TSA) is discernible in a given lesion or at least one type in combination with classical (unserrated) adenomatous tissue. The different histopathological types must be mentioned in the diagnosis, e.g. mixed polyp (HP and SSL, adenoma and SSL).

The vast majority of hyperplastic polyps and serrated lesions will not undergo malignant transformation. Only a fraction, especially in the group of sessile serrated lesions, may progress to rapidly aggressive carcinoma [35, 54].

Hyperplastic polyps rarely progress to carcinoma. A single case report can be found in the literature [55] and a second (unpublished) case has been reported in southern Germany. Interestingly, these carcinomas show features of gastric differentiation.

Little evidence is available on which the risk of colorectal cancer associated with serrated lesions other than hyperplastic polyps could be reliably judged. The risk assessment for sessile serrated lesions is not yet defined, but a subset of these lesions appears to give rise to carcinoma often less than a few millimeters in size. In a series of 110 traditional serrated adenomas, 37% exhibited foci of significant neoplasia and 11% contained areas of intramucosal carcinoma [12]. Mixed polyps (e.g., HP/TSA/SSL or HP/adenoma) seem to have at least the same rate of progression to colorectal carcinoma as adenomas, and the risk might be higher [56, 57].

Firstly, accurate measurement is very important, and measurement must be to the nearest millimeter (and not rounded-up to the nearest 5 or 10 mm). The maximum size of the lesion should be measured from the histological slide and if the lesion is disrupted or too large, from the formalin-fixed macroscopic specimen. If a biopsy is received it should be stated that size cannot be assessed.

Poorly differentiated carcinomas are identified by the presence of either irregularly folded, distorted and often small tubules, or the lack of any tubular formation and showing marked cytological pleomorphism. In the absence of good evidence, we recommend that a grade of poor differentiation should be applied in a pT1 cancer when ANY area of the lesion is considered to show poor differentiation. It should be noted that this is not in accordance with the WHO classification that recommends a certain proportion of lesion showing poor differentiation before diagnosing a lesion as G3. Poor differentiation includes undifferentiated and poorly differentiated as defined by the WHO classification [58].

Budding describes the biological behaviour of the tumour at the front of invasion [59]. Budding or tumour cell dissociation [60] can be divided into slight, moderate and marked and is known from the Japanese literature of the 1950s [61] and 1990s [62].

At this time, evidence is lacking concerning reproducibility of the numerous methods for tumour budding measurement (see Table 5). It is good practice but not mandatory to document the presence or absence of single tumour cells at the front of invasion and we therefore recommend providing this additional information in the written report with an explanatory comment as budding has been suggested as a prognostic factor in colorectal cancer [63‐65].

The site of origin of each specimen should be individually identified by the clinician and reported to the pathologist on the histopathology request form. The pathologist should record this on the proforma. This is important information because the risk of lymph node metastasis from a T1 adenocarcinoma varies depending on the site and size of the lesion (rectum vs. other locations) [66].

In columnar epithelium it is difficult to define the onset of invasive carcinoma and reliably distinguish it from high-grade intraepithelial neoplasia. Criteria such as single tumour cells are more likely to be seen in more advanced carcinomas, but not in early carcinomas. Desmoplastic stromal reactions are also seldom seen in very early carcinomas. However, basal membrane structures are frequently discernible in well-differentiated early carcinomas [67‐69], so that definitions using “invasion through the basement membrane” are incorrect.

The WHO definition of adenocarcinoma in use when the EU guidelines were developed excluded diagnosis of intramucosal carcinoma in the colon or rectum, in contrast to the accepted WHO definitions for the stomach, oesophagus and small bowel. In the latter cases, a decision on surgical versus local therapy is made based on respective protocols. Comparable lesions in the colon and rectum are reported as high-grade mucosal neoplasia, because a carcinoma in the colon is defined by infiltration of the submucosa, according to the WHO classification.

The discussion on this issue among the authors of the pathology chapter in the EU Guidelines reflects, among other things, concern about potential overtreatment of early T1 carcinomas which are detected much more frequently in a screening setting. The clinical management of a lesion where invasion of the lamina propria has occurred is no different from that where high-grade changes are confined to the glands. This legitimate concern as to increased morbidity and mortality due to miscommunication of diagnostic criteria may be dealt with more effectively in the future, as multidisciplinary management of lesions detected in and outside of screening programmes advances. The authors hope that such advances and their effective dissemination will be stimulated by the publication of the new EU guidelines. This, in turn, may lead to revision of the current WHO definition of colorectal adenocarcinoma in a future revision of the WHO classification of gastrointestinal tumours. Pathologists should report on what version of the WHO and TNM classifications their diagnosis is based.

In those cases in which intramucosal colorectal cancer is suspected, and particularly in countries in which this diagnosis is documented in addition to the WHO terminology, explicit comments by the pathologist are recommended. Based on the cytological characteristics of the case, the pathologist should indicate whether local endoscopic or surgical removal is recommended, and the basis for this recommendation should be indicated. This recommendation should be discussed in a multidisciplinary conference prior to surgery. The Japanese criteria for such stratification have been published [55]. The updated Paris classification based on a workshop in Feb 2008 in Kyoto [9] permits such subclassification based on improved grouping and explains in detail the grading criteria [10].

A subclassification of all carcinomas into low-risk and high-risk based on risk of lymph node involvement should always be undertaken. For exact criteria please see main text of the present publication and Chapter 7 of the full GL document [1] and the updated Paris classification [9, 10].

Perineural invasion (PNI) was recently described as an independent risk factor for colorectal cancer [66, 70]. PNI is significantly associated with high tumour stage, grade and metastases. Furthermore, PNI serves as an independent predictor of disease-free and cancer survival [66, 70]. Recently, an association with other criteria indicating an aggressive course of disease such as lymphatic vessel permeation, venous invasion, tumour growth pattern, budding [71] were described by Poeschl et al. [66]. Also, it was described that PNI-positive tumours are more likely to be incompletely resected, and more likely to progress after Mayo regimen chemotherapy than PNI-negative tumours. Lately, Poeschl et al. were able to show that PNI is an additional independent factor for local tumour relapse.

It is recommended to record PNI in routine sections of colorectal cancer. According to recent studies [66, 70, 72, 73], immunohistochemistry or special stains are not necessary to detect PNI. Prospective studies are needed to show the clinical relevance of PNI, its relationship to other features such as lymphatic and vascular invasion and the benefit of alternative treatment for such more aggressive tumours that are PNI positive.